







New research on that gives us a deeper understanding of why networks make the decisions they do
Neural Networks Are Already Clean
A working hypothesis about neurons, generalization, and the ecology of circuits.

Understanding Artificial Neural Networks: Mysterianism about Known Mechanism is Mysticism
Mysterianism is the idea that human cognition, mind, cannot be understood. Taking this concept and applying it to known mechanism — such that claims are made that we do not know how engineered systems, such as artificial neural networks (ANNs), work, or that they constitute black boxes that we can only open with difficulty — is inappropriate at best and malicious at worst. We do know the mechanistic structure of such models because we designed and built them. We also do know their functional role (what they are for) as well as the mathematical function they are asked to approximate (map inputs to target outputs). Because mysterianist beliefs about known systems, such as ANNs, are often expressed, scientists need to sit up and take notice. We provide an error theory as to what is going on to help unpick this metatheoretical blunder. Ultimately, the problem is that 'understanding' is not a technical term in these cases: the word is co-opted for a specific narrative to sell 'artificial intelligence' through mystification. All computational systems, from pendulums to databases, will behave in ways we cannot predict or control — this is not a unique property of ANNs — and experts do indeed grasp the computational properties of these systems nonetheless.
Understanding Artificial Neural Networks: Mysterianism about Known Mechanism is Mysticism
Mysterianism is the idea that human cognition, mind, cannot be understood. Taking this concept and applying it to known mechanism — such that claims are made that we do not know how engineered systems, such as artificial neural networks (ANNs), work, or that they constitute black boxes that we can only open with difficulty — is inappropriate at best and malicious at worst. We do know the mechanistic structure of such models because we designed and built them. We also do know their functional role (what they are for) as well as the mathematical function they are asked to approximate (map inputs to target outputs). Because mysterianist beliefs about known systems, such as ANNs, are often expressed, scientists need to sit up and take notice. We provide an error theory as to what is going on to help unpick this metatheoretical blunder. Ultimately, the problem is that 'understanding' is not a technical term in these cases: the word is co-opted for a specific narrative to sell 'artificial intelligence' through mystification. All computational systems, from pendulums to databases, will behave in ways we cannot predict or control — this is not a unique property of ANNs — and experts do indeed grasp the computational properties of these systems nonetheless.
OpenAI peeks into the “black box” of neural networks with new research
We do not understand" how LLMs work, admits OpenAI in quest to make them interpretable.

Distilling the Knowledge in a Neural Network
Our teams advance the state of the art through research, systems engineering, and collaboration across Google.

Artificial Intuition : The Improbable Deep Learning Revolution
This is an entirely unique book that was written in 2017. It is one of the very few that predicted today's Generative AI revolution. In this book, I wrote...
How neural networks learn distributed representations
Deep learning’s effectiveness is often attributed to the ability of neural networks to learn rich representations of data.

Understanding neural networks through sparse circuits
OpenAI is exploring mechanistic interpretability to understand how neural networks reason. Our new sparse model approach could make AI systems more transparent and support safer, more reliable behavior.

NeuroAI’s Next Steps: Balancing Computation and Embodiment
Deep neural networks trained at scale now predict neural activity across sensory cortices, raising a question that divides neuroscience: does the convergence between artificial and biological representations constitute understanding, or does understanding require models of biological processes grounded in the embodied, evolved character of natural intelligence? In this Dual Perspectives article, Bing Brunton and Patrick Mineault articulate two research programs—Process and Convergence—to engage this question about the future of NeuroAI research. The conversation is organized around four questions that probe what scaled model–brain alignment reveals, what the brain is coupled to and evolved to do, what success beyond the lab looks like, and where the field should invest over the coming decade. Their exchanges reveal the distance between the two programs to be primarily about the level of analysis and priorities for what to build next. The conversation converges on shared ground: measure broadly, then distill to causal principles.

Why I believe current AI models can really think and understand - Part 2 of 3
Why computers can in principle produce thinking and understanding

History of Neuroscience and AI: From neurons to the first artificial neural networks
Inceptionism: Going Deeper into Neural Networks
Posted by Alexander Mordvintsev, Software Engineer, Christopher Olah, Software Engineering Intern and Mike Tyka, Software EngineerUpdate - 13/07/20...

On Network Science and Mutual Information for Explaining Deep Neural Networks
View recent discussion. Abstract: In this paper, we present a new approach to interpret deep learning models. By coupling mutual information with network science, we explore how information flows through feedforward networks. We show that efficiently approximating mutual information allows us to create an information measure that quantifies how much information flows between any two neurons of a deep learning model. To that end, we propose NIF, Neural Information Flow, a technique for codifying information flow that exposes deep learning model internals and provides feature attributions.
Representer Point Selection for Explaining Deep Neural Networks
We propose to explain the predictions of a deep neural network, by pointing to the set of what we call representer points in the training set, for a given test point prediction. Specifically, we show that we can decompose the pre-activation prediction of a neural network into a linear combination of activations of training points, with the weights corresponding to what we call representer values, which thus capture the importance of that training point on the learned parameters of the network. But it provides a deeper understanding of the network than simply training point influence: with positive representer values corresponding to excitatory training points, and negative values corresponding to inhibitory points, which as we show provides considerably more insight. Our method is also much more scalable, allowing for real-time feedback in a manner not feasible with influence functions.
Artificial Intuition — A Breakthrough Cognitive Paradigm
In a previous post, I introduced the Meta Meta-Model of Deep Learning. However, I did not introduce its details. A word of warning for the…

Position: Don't Just "Fix it in Post": A Science of AI Must Study Training Dynamics
What would it mean to have a scientific understanding of AI? Models are not static objects: they are snapshots of time-evolving processes shaped by data, objectives, architectures, and optimization dynamics. Yet much of AI research treats models as fixed artifacts, analyzing behaviors after training rather than asking why they emerge. This position paper argues that a science of AI must move beyond post-hoc fixes and study the training dynamics that produce model behavior. Such a science should support progressively stronger forms of understanding: predicting outcomes from early training signals, intervening when trajectories go wrong, and ultimately designing training procedures that more reliably produce desired properties. Scaling laws have made prediction routine for loss; the challenge is extending this success to capabilities, biases, robustness, and safety-relevant behaviors. We articulate requirements for such theories grounded in the history and philosophy of science, examine progress in mechanistic interpretability, fairness, memorization, and simplicity bias, and identify concrete open problems.
